NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Segal JB, Strouse JJ, Beach MC, et al. Hydroxyurea for the Treatment of Sickle Cell Disease. Rockville (MD): Agency for Healthcare Research and Quality (US); 2008 Feb. (Evidence Reports/Technology Assessments, No. 165.)

  • This publication is provided for historical reference only and the information may be out of date.

This publication is provided for historical reference only and the information may be out of date.

Cover of Hydroxyurea for the Treatment of Sickle Cell Disease

Hydroxyurea for the Treatment of Sickle Cell Disease.

Show details


The objective of the report is to review and synthesize the available evidence regarding the efficacy and effectiveness of hydroxyurea treatment in patients with sickle cell disease, to assess the potential short and long-term harms of its use in patients with sickle cell disease and other diseases, and to discuss barriers to the use of hydroxyurea and other medications in the treatment of sickle cell disease. The results of this report will be presented to an NIH Consensus Panel in February 2008.

Recruitment of Technical Experts and Peer Reviewers

We assembled a core team of experts from JHU who have strong expertise in the management of and research in sickle cell disease, clinical trial methodology (including clinical trials of hematological agents), systematic literature review, epidemiological studies, and ethics and adherence research. We also recruited external technical experts from diverse professional backgrounds, including academic, clinical, and non-profit public interest groups. The core team asked the technical experts for input regarding key steps of the process, including the selection and refinement of the questions to be examined. Peer reviewers were recruited from various clinical settings. Bristol-Myers Squibb, maker of Droxia® and Hydrea®, was invited to review the draft report and declined in writing. In addition to Bristol-Myers Squibb, eight generic manufacturers of hydroxyurea were invited to serve as reviewers. The eight manufacturers declined in writing, were no longer manufacturing hydroxyurea, or did not reply to two or more written requests. (See Appendix F *.)

Key Questions

The core team worked with the technical experts, the OMAR Consensus Panel chairman, and the AHRQ to develop the Key Questions that are presented in the “The Purpose of This Evidence Report” section of Chapter 1 (Introduction). Before searching for the relevant literature, we clarified our definitions of these Key Questions and the types of evidence that we would include in our review.

Key Questions 1 and 2 addressed the efficacy (the therapeutic effect of an intervention in an ideal setting, such as a clinical trial) and effectiveness (the therapeutic effect of an intervention as demonstrated or observed in patients in their usual care setting) of hydroxyurea in patients with sickle cell disease. Based on discussion with our experts, we knew that limiting our search to randomized trials would yield an insufficient number of articles upon which to draw conclusions. Therefore, we opted to include RCTs, cohort studies with a control population, and pre/post studies. We planned to address efficacy outcomes in both children and adults. We chose not to include case series in our review of efficacy and effectiveness, since these studies would not yield strong evidence for efficacy. We opted to include studies of biomarkers as intermediary indicators of efficacy if they were of the appropriate study design (RCTs, controlled cohort studies, or pre/post studies) (Figure 1).

Figure 1. Analytic Framework.


Figure 1. Analytic Framework.

Key Question 3 addressed the toxicity of hydroxyurea in patients with sickle cell disease. To respond to this question, we chose to look for strong evidence of toxicity in patients with sickle cell disease by reviewing controlled studies (randomized, non-randomized, and pre/post studies) that had addressed toxicities in this population. Given that the CERHR26 has recently reported in detail on toxicities to children and developing fetuses, we chose to update and confirm the findings presented in that report without producing our own detailed description of the developmental toxicities of hydroxyurea in children and fetuses.

Since we anticipated that the availability of strong evidence would be limited, we chose to also allow weaker forms of evidence such as case reports. We decided to exclude case series, since the level of detail in reports of cases series is generally insufficient to allow us to assess how the outcome is causally related to the exposure. To provide further information regarding the potential toxicities of this drug, we chose to also include indirect evidence of any toxicity in other patient populations treated with hydroxyurea. As noted above, we chose to include strong evidence of toxicities in other patient populations by reviewing controlled studies (both randomized and non-randomized and pre/post studies). We also included case reports in these populations, but not case series. The exception was the few very large case series (100 or more patients) reporting toxicities in patients with diseases other than sickle cell disease, excluding CML. Since we found no other source of published information on long-term exposure to hydroxyurea, we reasoned that these studies might provide useful, although indirect, evidence of particular toxicities.

Key Question 4 concerned barriers to the use of hydroxyurea. We anticipated finding little in the way of data that specifically addressed barriers to the use of this drug for sickle cell disease. Therefore, we sought information on barriers to the use of other therapies for treatment of sickle cell disease, including the receipt of routine, scheduled care; adherence to medications; and receipt of therapies, including pain control and prescriptions. We hypothesized that these barriers would be representative of barriers to the use of hydroxyurea. We opted to search for: (1) studies that tested whether supposed barriers were actual barriers to accessing scheduled care, receiving medication prescriptions, or adhering to medications; (2) studies in which patients, providers, or family members described what they perceived to be barriers to accessing scheduled care, receiving medication prescriptions, or adhering to medications; and (3) studies that tested an intervention aimed at overcoming barriers to accessing scheduled care, receiving medication prescriptions, or adhering to medications (Figure 2).

Figure 2. Conceptual Model: Hydroxyurea Treatment For Sickle Cell Disease (Adapted From The Aday & Andersen Expanded Behavioral Model).


Figure 2. Conceptual Model: Hydroxyurea Treatment For Sickle Cell Disease (Adapted From The Aday & Andersen Expanded Behavioral Model).

Literature Search Methods

Searching the literature involved identifying reference sources, formulating a search strategy for each source, and executing and documenting each search. For the searching of electronic databases, we used medical subject heading (MeSH) terms that were relevant to hydroxyurea, combined with sickle cell disease and with other hematologic diseases such as essential thrombocythemia. We used a systematic approach to searching the literature in order to minimize the risk of bias in selecting articles for inclusion in the review.

This strategy was used to identify all the relevant literature that applied to our Key Questions. We also looked for eligible studies by reviewing the references in pertinent reviews, by querying our experts, and by taking advantage of knowledge shared at core team meetings.


Our comprehensive search included electronic and hand searching. On March 15, 2007, we ran searches of the MEDLINE® and EMBASE® databases. A supplemental search targeting essential thrombocythemia was added to the MEDLINE and EMBASE searches on May 7, 2007. On June 30, 2007, the MEDLINE and EMBASE searches were updated, and additional searches were executed using TOXLine and CINAHL. All searches were limited to English-language articles involving treatment of humans. Review articles were excluded from the searches. Searches were not limited by date of publication or by subject age.

Search Terms and Strategies

Search strategies specific to each database were designed to enable the team to focus the available resources on articles that were most likely to be relevant to the Key Questions. We developed a core strategy for MEDLINE, accessed via PubMed, on the basis of an analysis of the MeSH terms and text words of key articles identified a priori. The PubMed strategy formed the basis for the strategies developed for the other electronic databases (see Appendix A *).

Organization and Tracking of the Literature Search

The results of the searches were downloaded into ProCite® version 5.0.3 (ISI ResearchSoft, Carlsbad, CA). Duplicate articles retrieved from the multiple databases were removed prior to initiating the review. From ProCite, the articles were uploaded to SRS 4.0 (TrialStat © 2003-2007). SRS is a secure, Web-based collaboration and management system designed to speed the review process and introduce better process control and scientific rigor. We used this database to store full articles in portable document format (PDF) and to track the search results at the title review, abstract review, article inclusion/exclusion, and data abstraction levels.

Title Review

The study team scanned all the titles retrieved. Two independent reviewers conducted title scans in a parallel fashion. For a title to be eliminated at this level, both reviewers had to indicate that it was ineligible. If the first reviewer marked a title as eligible, it was promoted to the next elimination level. If the two reviewers did not agree on the eligibility of an article, it was automatically promoted to the next level (see Appendix B, Title Review Form).

The title review phase was designed to capture as many studies as possible that reported on the efficacy and/or effectiveness of hydroxyurea treatment of hematologic diseases, the toxicity of hydroxyurea in the treatment of any disease, and the barriers to the treatment of sickle cell disease with hydroxyurea or other agents. All titles that were thought to address the above criteria were promoted to the abstract review phase.

Abstract Review

The abstract review phase was designed to identify articles that applied to our Key Questions. An abstract was excluded at this level if it did not apply to the Key Questions or for any of the following reasons: It was not written in English, contained no original data, involved animals only, was solely a report of an in vitro experiment, or was a case series of fewer than 10 patients, unless the article was primarily reporting on toxicities (Appendix B *, Abstract Review Form).

Abstracts were promoted to the article review level if both reviewers agreed that the abstract could apply to one or more of the Key Questions and did not meet any of the exclusion criteria. Differences of opinion were resolved by discussion between the two reviewers.

Article Review

Full articles selected for review during the abstract review phase underwent another independent review by paired investigators to determine whether they should be included in the full data abstraction. At this phase of review, investigators determined which of the Key Question(s) each article addressed (see Appendix B, Article Inclusion/Exclusion Form). If articles were deemed to have applicable information, they were included in the data abstraction. Differences of opinion regarding article eligibility were resolved through consensus adjudication.

Once an article was included at this level, an additional level (filter) was added to further exclude articles that were found to be inapplicable once the data abstraction was underway. This process was used to eliminate articles that did not contribute to the evidence under review (see Appendix B, Triage Form). Articles could be excluded at this level for the following reasons: They contained insufficient data to address the question or only a very minimal description of a study population (e.g., they provided no relevant outcome data, no details about the included patients, or no description about the intervention except that it involved hydroxyurea). We excluded studies with fewer than 20 patients unless the article was primarily reporting on the toxicity of hydroxyurea in sickle cell disease. We excluded trials involving diseases other than sickle cell disease if fewer than 20 patients received hydroxyurea. We allowed case series if they described toxicities in more than 100 patients. We excluded case reports if there was no description of duration of use of hydroxyurea or no description of the dose(s) used, or if the study addressed pregnancy. A list of the articles excluded at this level is included in Appendix D.

Data Abstraction

After applying the criteria described above, we used a sequential review process to abstract data from the remaining articles. In this process, the primary reviewer completed all the relevant data abstraction forms. The second reviewer checked the first reviewer's data abstraction forms for completeness and accuracy. Reviewer pairs were formed to include personnel with both clinical and methodological expertise. The reviews were not blinded in terms of the articles' authors, institutions, or journal. 27 Differences of opinion that could not be resolved between the reviewers were resolved through consensus adjudication.

For all articles, excluding case reports, reviewers extracted information on general study characteristics: study design, location, disease of interest, inclusion and exclusion criteria, and description of administered therapies (see Appendix B *, General Form). Participant characteristics were also abstracted: information on intervention arms, age, race, genotype and haplotype, substance abuse, socioeconomic status, and related data on the disease under study.

Outcome data were abstracted from the articles that were applicable to the Key Questions regarding hydroxyurea's efficacy and/or effectiveness and its toxicity. Reviewers abstracted data on both categorical and clinical outcomes and toxicities (see Appendix B, Key Questions 1–3). Case reports on hydroxyurea toxicity were abstracted using a separate form. The reviewers abstracted data on disease, subject age, the reported adverse event(s), and causality using the WHO's causality assessment instrument described below28 (see Appendix B, CR Tox).

Separate forms were developed to abstract data for Key Question 4 (see Appendix B, Key Question 4 Form). For each study, we determined the extent to which the measured study outcomes were likely to be true measures of the outcome of interest (e.g., provision of appropriate pain management or receipt of routine, scheduled care). For example, in the pain management interventions, we considered utilization outcomes (e.g., hospital length of stay or costs) and descriptive comments from patients (without explicit qualitative methodology to analyze those comments) to be forms of indirect evidence, and we considered variables abstracted by chart review (e.g., ratings of patient-controlled analgesia, pain consults, or patient pain ratings) to be forms of direct evidence.

For Key Question 4, we categorized each study as providing “direct” or “indirect” evidence. Studies in which there was at least one outcome that was considered to be a true measure of our outcome of interest were considered to provide “direct” evidence. We categorized the study as providing “indirect” evidence if either (1) only indirect outcomes were measured or (2) both direct and indirect outcomes were measured, but only the indirect (and not the direct) outcome demonstrated an effect.

For each study designed to test interventions to overcome treatment barriers, we determined by consensus of two reviewers whether there was “improvement,” “partial improvement,” “no improvement” or a “detrimental” effect. We categorized intervention studies as indicating “improvement” if some, most, or all measured outcomes showed statistically significant improvement and no outcomes worsened. We categorized intervention studies as indicating “potential improvement” if the authors implied that some, most, or all measured outcomes had improved and they gave data to suggest that their conclusions were correct but did not perform statistical tests. We categorized intervention studies as indicating “partial improvement” if our main outcome of interest did not improve as a result of the intervention, but there were other positive effects. We categorized intervention studies as showing “no improvement” if there was no improvement in any outcome and no outcomes worsened. We categorized intervention studies as “detrimental” if some, most, or all measured outcomes worsened and no outcomes improved.

Quality Assessment

We assessed the included studies on the basis of the quality of their reporting of relevant data. For the randomized controlled trials, we used the scoring system developed by Jadad et al. 29: (1) Was the study described as randomized (this includes the use of words such as “randomly,” “random,” and “randomization”)? (2) Was the method used to generate the sequence of randomization described, and was it appropriate? (3) Was the study described as double-blind? (4) Was the method of double-blinding described, and was it appropriate? (5) Was there a description of withdrawals and dropouts?

For the observational studies (both cohort studies and controlled clinical trials), we created a quality form based on those previously used by our EPC. This form was aimed primarily at capturing data elements most relevant to study design. We designed questions to evaluate the potential for selection bias, which might limit internal validity and generalizability, as well as questions to assess the potential for confounding, which could bias the estimates of the treatment effect. 3032 For our assessment of the quality of the qualitative studies we reviewed, we developed a form to identify key elements that should be reported when describing the results of qualitative research, as advocated by leaders in the field. 3335 For our quality assessment of the surveys reviewed, we adapted information from Ratanawongsa et al. 36 The quality assessments were done independently by paired reviewers. A third reviewer reconciled the results of the first two reviewers in the case of the randomized trials. 29 For the other study designs, the results of the two reviewers were averaged. The quality assessment instruments are included in Appendix B, Quality Forms.

Data Synthesis

We created a set of detailed evidence tables containing information extracted from the eligible studies. We stratified the tables according to the applicable Key Question(s). Once evidence tables were created, we re-checked selected data elements against the original articles. If there was a discrepancy between the data abstracted and the data appearing in the article, this discrepancy was brought to the attention of the investigator in charge of the specific data set, and the data were corrected in the final evidence tables.

We did not quantitatively pool the data for any of the outcomes because there was a paucity of RCTs addressing any of our outcomes of interest. The substantial qualitative heterogeneity among the observational studies (with different populations, different dosage schedules, and different durations of follow-up) made pooling these studies inadvisable.

Data Entry and Quality Control

Data were abstracted by one investigator and entered into the online data abstraction forms (see Appendix B, Forms). Second reviewers were generally more experienced members of the research team, and one of their main priorities was to check the quality and consistency of the first reviewers' answers.

Grading of the Evidence

At the completion of our review, we graded the quantity, quality, and consistency of the best available evidence, addressing Key Questions 1 and 2 together and Key Question 3 alone, by adapting an evidence grading scheme recommended by the GRADE Working Group37 and modified in Chapter 11 of the EPC Manual currently under development. We separately considered the evidence from studies of children and studies of adults. In rating the strength of the study designs, RCTs were considered to be best, followed by non-RCTs and observational studies. If an outcome was evaluated by at least two RCTs as well as observational studies and case reports, our evidence grade was based only on the RCTs evaluating that outcome. If an outcome was evaluated by one or no RCTs, our evidence grade was based on the single RCT (if any) in addition to the best available non-RCT or the best available observational studies (cohort studies considered best, followed by cross-sectional studies and studies with a pre/post observational design). The results of case reports were incorporated into the grading of Key Question 3 as described below.

We assessed the quality and consistency of the best available evidence, including an assessment of the risk of bias in relevant studies (using individual study quality scores), whether the study data directly addressed the Key Questions, and the precision and strength of the findings of individual studies. We classified evidence bodies pertaining to each Key Question into four basic categories: (1) “high” grade (high confidence that the evidence reflected the true effect; further research is very unlikely to change our confidence in the estimate of the effect); (2) “moderate” grade (moderate confidence that the evidence reflected the true effect; further research may change our confidence in the estimate of effect and may change the estimate); (3) “low” grade (low confidence that the evidence reflected the true effect; further research is likely to change the confidence in the estimate of effect and is likely to change the estimate); and (4) “insufficient” (evidence was either unavailable or did not permit the estimation of an effect) (Appendix E).

The evidence regarding the case reports was graded according to the WHO Collaborating Center for Drug Monitoring. 28, 38 A reaction was rated as “certain” if all four criteria for causality were fulfilled: (1) a plausible time relationship between drug administration and an event; (2) an absence of a concurrent disease that might have caused the event; (3) a reasonable response to drug withdrawal; and (4) existence of a rechallenge or a demonstrated biological explanation. A reaction was rated as “probable” if criteria 1, 2, and 3 were fulfilled, and “possible” if only criterion 1 was met and information on criterion 3 was lacking or unclear. A reaction was rated as “unlikely” if criterion 1 was not met and if other drugs, chemicals, or underlying disease provided a plausible explanation for the reaction. We rated a reaction as “possible” if only criterion 1 was met and the reaction did not meet criteria for “certain.” After these causality assessments, we assigned a level of evidence to each reported potential adverse event: Level 1 evidence had to have at least one certain case report, level 2 evidence had to have at least one probable report but no certain report, and level 3 evidence had to have at least one possible report but no certain or probable case report. The level 1 evidence was used as supportive evidence when assigning an evidence grade to the whole body of evidence for Key Question 3.

We graded the evidence for Key Question 4 using two instruments: The sub-question regarding interventions to overcome barriers was graded using the instrument described above. We graded the evidence regarding the existence of barriers using a modification of this instrument that addressed similar domains: the quantity of studies, protection against bias in the studies (quality), and consistency (Appendix E).

For each outcome of interest, two investigators graded each Key Question, and then the entire team discussed their recommendations and reached a consensus.

Peer Review

Throughout the project, the core team sought feedback from the external technical experts and the OMAR panel. A draft of the report was sent to the technical experts and peer reviewers, as well as to representatives of the AHRQ and the NIH (OMAR). In response to the comments from the technical experts and peer reviewers, we revised the evidence report and prepared a summary of the comments and their disposition that was submitted to the AHRQ.



Appendixes cited in this report are provided electronically at: http://www​​.htm


  • PubReader
  • Print View
  • Cite this Page

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...